An Adaptive Mechanism for Web Browser Cache Management

R esearchers have applied various caching techniques to decrease the network loads and response times caused by the phenomenal growth of the World Wide Web. These strategies include the use of callback, prefetching, and validation.1 However, callback mechanisms, which were developed for dedicated distributed systems, are not appropriate for Web objects, which may be cached in many proxies. Prefetching is also unsuitable, as current cache hit rates are typically only about 50 percent.2 It is, therefore, difficult to know which objects to prefetch or when to prefetch them, and thus impossible to ensure that preemptive document checking will improve cache performance. By contrast, validation has been implemented in most proxy servers and browsers. With this technique, cached files are time stamped with an expiry date, which allows the cached version’s currency to be checked when future requests are made. However, many browsers ignore the document expiry, potentially setting up a coherence problem, where stale documents are served as current.3 The validation method entails three decisions that influence effective cache management: